Byakugan: Private, open-source AI-text checker
AI Checker
Check whether writing reads as AI, on your own machine, then improve how it reads. Free and open source.
Two ways to use it.
1. In your browser
A static web app. Paste your text and get a read on how AI-like it sounds, with a plain-language explanation, a grammar pass, and a check against your own documents for reuse. Everything runs in the page, and nothing you paste leaves the browser.
Run it locally:
cd site
npm install
npm run devRelated MCP server: AI Content Detector MCP Server
2. Inside the AI you already use
A mcp server for your AI assistant (Claude Desktop, an IDE, any MCP client). It gives that AI a detection read plus the exact phrases that read as AI, so your own AI can check and rewrite your writing, on your machine.
For Claude Desktop, add this to claude_desktop_config.json and restart:
{
"mcpServers": {
"ai-checker": {
"command": "npx",
"args": ["-y", "@jayofemi/ai-checker"]
}
}
}Claude Code users can install it as a plugin instead: /plugin marketplace add JayOfemi/byakugan, then /plugin install ai-checker@ai-checker.
Then tell your AI "use ai-checker to check this text" and paste what you want checked. Full setup, the tools, and how to verify it are in mcp/README.md.
What to expect
AI-text detection is a signal, not a verdict, and it is biased against non-native English writers. It is shown as a likelihood with its confidence, never a yes/no judgement. Stiff, even, stock-phrase-heavy writing reads high; varied, personal writing reads low. Use it on your own work.
Layout
site/- the browser checker, a static web app. See site/README.md.mcp/- the MCP server and TypeScript library. See mcp/README.md.
Each folder is its own package with its own build. Node 22 (.nvmrc).
License
MIT. Copyright (c) 2026 Jay Ofemi.
Available Tools
4 toolscheck_grammarCheck grammarA
Run a rule-based grammar and style check over the text, fully on-device, and return each issue with its character span, a message, the flagged text, and suggested fixes. Use it to tidy grammar and phrasing, before or after rewriting the flagged AI tells. Deterministic.
Inputs: text (required).
check_reuseCheck reuse overlapA
Compare the text against reference documents you provide, using word shingles, and return the share of overlap and the verbatim shared passages. This is a local check against the documents passed in. It does not search the web or any plagiarism database. Deterministic.
Inputs: text (required), references (required).
detect_ai_textDetect AI textA
Score how likely a passage reads as AI-written, fully on-device, with a per-signal breakdown and a confidence band. Returns a 0 to 100 likelihood, never a yes/no verdict. The signal is directional and biased against non-native English writers, so treat it as a second opinion and not proof. Get a read here, then call find_ai_tells to locate the spans worth rewriting.
Inputs: text (required).
find_ai_tellsFind AI tellsA
Locate the exact character spans that read as AI, covering stock phrasing, long dashes, and formulaic sentence openers. Returns a list of spans with start, end, the matched text, and a note. Use these to rephrase only the flagged parts and leave the rest of the writing intact. Deterministic.
Inputs: text (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 4 tools
Each tool has a clearly distinct purpose: grammar checking, reuse detection, global AI-likelihood scoring, and localized span detection. No overlap in functionality.
All tool names follow a consistent verb_noun snake_case pattern (check_grammar, check_reuse, detect_ai_text, find_ai_tells) with uniform verb usage.
Four tools is an appropriate scope for a text checker covering grammar, reuse, and two distinct AI-detection features. Not too few or too many.
The tool set covers grammar, reuse, and AI detection comprehensively. A minor gap is absence of a rewrite/repair tool, but the purpose is detection, not editing.
Maintenance
Related MCP Connectors
Find AI-isms with evidence and fingerprint a writing voice from samples. 3 of 5 free.
Paste a draft and see the habits that make writing read like it came from an AI model.
Detect AI slop in text, HTML, pages, and project folders. Free, no API key.
Scores whether a page reads as AI-generated and names the tells to fix.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceAI text detection and humanization for Claude Code, Cursor, and Windsurf. Check if text sounds AI-generated, improve it to read naturally, and verify results — all without leaving your editor.46 npmMIT
- FlicenseNot gradedqualityCmaintenanceDetects AI-generated content by analyzing 24 writing patterns, provides style analysis, plagiarism check, text humanization suggestions, and offers batch detection and version comparison tools.-
- AlicenseAqualityBmaintenanceBilingual (EN/ES) AI-writing detection that shows the evidence instead of a percentage: named tells with line and column, hidden-character inspection, and citation cross-checking against a document's own bibliography. Seven of its nine tools run entirely locally and never touch the network.1031MIT
- AlicenseAqualityBmaintenanceProvides local, deterministic scoring and detailed auditing of text for AI-writing patterns via two read-only tools, with no network calls or language models.29 npm3MIT